Specific Aim 1: To improve the understanding of the role of the MET pathway in patients with advanced urothelial carcinoma Tumor Tissue MET expression: To describe the percentage of HGF and MET overexpression in urothelial tumors: [1] Formalin-fixed paraffin-embedded (FFPE) primary tumor samples from previously treated patients on other protocols (n=68) and primary tumors from cabozantinib study patients (n=55) will be analyzed for MET and HGF expression by IHC. Pre-cabozantinib treatment tumor biopsies of metastatic sites will also be obtained (so far 50% of patients agreed to a biopsy) and baseline MET and HGF expression of primary and metastatic tumor sites of these patients will be compared. Post treatment biopsies will be performed on a voluntary basis, to enable pre versus post comparisons of MET and HGF levels. [2] In fresh frozen tissue (FFT), HGF, MET, phosphoMET, RET, KIT, AXL and FLT3 will be quantitatively analyzed using a two-site electrochemiluminescent immunoassay developed for use with a Meso Scale Discovery (MSD) SectorImager 2400 plate reader in tumor samples from previously treated patients at the NCI (n=20) and at baseline from patients on study with cabozantinib (n=25). Use of a purified recombinant MET ectodomain-Ig fusion protein (358-MT, R&D Systems) as a reference standard permits quantitation as MET mass per mass total extracted cell protein. HGF, RET, KIT, AXL and FLT3 proteins will be quantitated similarly. Post treatment biopsies will be performed on a voluntary basis, and HGF, MET, phosphoMet, RET, KIT, AXL and FLT3 expression levels will be compared to baseline measurements. MET and HGF expression by IHC will be compared to MET, phosphoMET, and HGF expression by quantitative two-site electrochemiluminescent immunoassay. Results will be correlated with baseline prognostic factors such as tumor grade, stage, size, sites of metastatic disease, response to prior cisplatin therapy, response to cabozantinib, progression free survival and overall survival. Plasma MET expression: To determine whether plasma HGF, MET, VEGF-A, sVEGFR2 and PlGF levels analyzed by electrochemiluminescent immunoassay at baseline and after cabozantinib treatment are prognostic and/or predictive biomarkers of response to systemic therapy. Results will be correlated with baseline prognostic factors, response to cabozantinib, progression free survival and overall survival. Urine MET expression: To determine whether urinary HGF and soluble MET receptor (sMET) at baseline and after cabozantinib treatment are prognostic and/or predictive biomarkers of response to systemic therapy. Results will be correlated with baseline prognostic factors, source of urine (intact bladder, conduit, neobladder or nephrostomy), response to cabozantinib, progression free survival and overall survival. Genomic DNA analysis: Genes that are important downstream effectors of several tyrosine kinases involved in urothelial cancer and cabozantinib's targets will be analyzed and correlated with response. Genomic DNA will be prepared from FFPE. Tumor samples will be analyzed for MET, MET amplification, RET, PTEN, PIK3CA, and PIK3R1. AIM 2: To improve the outcome of patients with advanced urothelial carcinoma through the use of cabozantinib, a multi-receptor tyrosine kinase, primarily targeting MET and VEGFR2. A Phase II Study of Cabozantinib in Patients with Advanced/Metastatic Urothelial Carcinoma: we will formally test the hypothesis that cabozantinib, an oral dual-receptor tyrosine kinase inhibitor of MET and VEGFR2, improves the objective response rate in the second-line setting by 50-100%, i.e. from 10-15% (consistently observed with conventional chemotherapy) to at least 20%. We will also assess progression-free and overall survival. This open label, non-randomized, phase II trial of cabozantinib, 60 mg/day has been approved by CTEP and the NCI IRB (NCT01688999). Each patient will undergo response evaluation assessments with a CT scan of the chest, abdomen and pelvis at baseline and every 8 weeks (2 cycles) using RECIST 1.1 criteria by a central reference radiologist. Patients will remain on study until the time of disease progression or the development of unacceptable toxicity that is not manageable with dose reduction. Eligible patients must have a diagnosis of metastatic, progressive urothelial carcinoma of the bladder, urethra, ureter, or renal pelvis, and must have received at least one prior cytotoxic chemotherapy regimen. Patient must be able to provide either archival tumor samples (H&E slides and one paraffin block or 10 unstained slides) and/or undergo tumor biopsy in order to enroll on study. The study follows an optimal two-stage design, in order to rule out an unacceptably low 5% overall response rate (ORR; p0=0.05), in favor of a modest response rate of 20% (p1=0.20). With alpha=0.05 and beta = 0.10, the study will initially enroll 21 evaluable patients and if 0 to 1 of the 21 respond by RECIST version 1.1, then no further patients will be accrued. If 2 or more of the first 21 respond, then accrual would continue. A maximum of 55 subjects will be enrolled. Up to 45 patients will be accrued to cohort 1 (metastatic, progressive urothelial cancer). The remainder will be enrolled on exploratory cohorts 2 & 3, bone only metastatic urothelial disease and non TCC bladder cancer, respectively, during the time the study is accruing patients to cohort 1.